KR20150046431A - Auto-learning system and method for derive effective marketing - Google Patents

Abstract

The present invention relates to an auto-learning system and method for deriving effective marketing, by which bulletin board messages uploaded in an SNS are analyzed to be utilized in the marketing, and to an auto-learning system through which a better marketing is derived by comparing SNS bulletin board messages and writer information in a conventional marketing way, the derived contents are stored to be utilized as reference data in a next marketing process, so that an optimal marketing way can be derived according to a business field and a situation.

Description

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an automatic learning system and method,

The present invention relates to an automatic learning system and method for extracting effective marketing that is used for marketing by analyzing posts uploaded to the SNS. It compares information of SNS posts and authors with existing marketing methods to derive better marketing, The present invention relates to an automatic learning system capable of deriving an optimum marketing method according to the type of business or situation by storing contents and using it as a reference material for next marketing.

As Internet technology develops, online services are being offered in various fields such as culture, shopping, entertainment, and marketing. Among them, online marketing has been started in a form of unilaterally communicating information of a company or product to users through the first homepage. Recently, various marketing using SNS has been attempted as the use of SNS has increased rapidly.

SNS is a service that helps to connect and communicate with unspecified people online, and as the introduction of high-performance terminal devices such as smart phones and wireless networks that are inexpensive to use such as Wi-Fi networks, There is a trend.

In particular, the SNS is a one-to-one direct relationship between SNS users and information exchange among SNS users who have a relationship, and SNS users having similar interests are likely to form a mutual relationship and form a community. There is a high tendency to trust information shared among users.

In applying this marketing to the SNS, it is expected that the marketing effect will be maximized if the SNS user or a plurality of SNS user groups are interested and provided with matching marketing information that is likely to spread.

Accordingly, the present invention has been made in view of the above-mentioned matters, and it is an object of the present invention to search for articles posted on the SNS through keywords, to detect the objects to be promoted, and to derive public relations topics and methods suitable for the objects to be promoted.

In order to achieve the above object, an automatic learning system and method for deriving effective marketing according to the technical idea of the present invention includes a keyword registration unit for setting and registering a search keyword, a search unit for searching a post including the search keyword from the SNS, An SNS collector for collecting information; an information extracting unit for extracting information and author information necessary for deriving a publicity method from the retrieved post information; and an information extracting unit for extracting post information based on the post information extracted by the information extracting unit, An offer derivation unit for selecting a publicity subject based on information of the post information and the post creator extracted by the information extracting unit; The post information and the information of the post creator are stored in the learning storage unit An offer execution unit for performing a publicity by a publicity method derived by the channel selection unit from a publicity subject selected by the offer derivation unit to a publicity target selected by the target selection unit; A result collecting unit collecting information of a publicity target in response to the publicity performed by the offer executing unit; a result processing unit delivering the publicity target information collected in the response collecting unit to the learning storing unit; And a learning storage unit in which a subject, a publicity method, and a publicity response are stored.

In addition, the information extracting unit may evaluate the interest based on at least one of a publication date, a number of posts created during an arbitrary period, and a creation amount when collecting post information including a search keyword and information of a post creator can do.

The information extracting unit may calculate the degree of influence based on the number of other users related to the author in the SNS when collecting the post information including the search keyword and the information of the post creator.

If there is a group having the same characteristics as those of the past public relations object stored in the learning storage unit, the target selection unit may add the group immediately as a publicity target or give a contribution point when determining whether to select the group as a publicity target And if it does not have the same characteristics as the group of past publicity targets of the learning storage unit, it is determined whether or not to select a publicity target based on only information received from the information extraction unit.

The offer derivation unit may select a publicity subject by reflecting the response rate when publicity is performed on the publicity subject if the publicity subject performed in the current environment is already stored in the learning storage unit when the publicity subject is selected . ≪ / RTI >

Further, the offer execution unit may perform a publicity by placing a difference in the environment or the time when the publicity is executed.

The response collecting unit may collect the response time, the time until the response after the publicity run, and the environment information when the publicity target has responded to the publicity performed by the offer execution unit.

Further, the offer execution unit may perform a promotion by combining two or more promoting methods in the same environment or at the same time.

The reaction collecting unit may calculate the degree of synergy that occurs when the degree of the reaction rate is higher than that of the single public relations method and the different types of public relations methods are combined when the public relations object is answered with the combined public relations performed by the offer execution unit .

According to the automatic learning system and method for deriving effective marketing as described above, there is the following effect.

First, it is possible to grasp whether the SNS user's interest and the interest are positive or negative by keyword search performed by the collection unit.

Second, the process of comparing the current group and the past group stored in the learning storage unit is repeated, and as the PR is performed to the group, the PR result value is stored in the continuous learning storage unit so that the SNS users are more precisely grouped .

Third, as the results of the offer derivation section are derived from various public relations topics and the results of the public relations promotion are accumulated in the learning storage section, the public relations topic having the best effect can be easily found.

Fourth, as the number of times the PR method derived by the channel selector increases, the degree of effectiveness of the PR methods can be grasped in correspondence with the PR object and the PR topic, thereby facilitating the most effective and cheap PR method .

Fifth, when the compound advertisement method derived by the channel selector is used for the actual advertisement and the information about the reaction rate of the advertisement is accumulated, it is possible to easily find the advertisement methods having the synergy effect at the same time.

Sixth, when the offer executor executes the public relations, if the publicity is performed with a difference in the environment and the results are stored in the learning storage unit, it is possible to grasp how the public relations topic and the public relations method are affected by the environment.

1 is a block diagram illustrating a configuration of an automatic learning system and method for deriving effective marketing according to an embodiment of the present invention;
2 is a message flow diagram illustrating a service process of an automatic learning system and method for deriving effective marketing according to an embodiment of the present invention.
FIG. 3 is a flowchart illustrating an operation sequence of an automatic learning system and method for deriving effective marketing according to an embodiment of the present invention.
4 is a flowchart for explaining driving of a target selection unit in an embodiment of the present invention.
5 is a flowchart for explaining driving of an offer derivation unit in an embodiment of the present invention.
6 is a flowchart for explaining the operation of the channel selection unit in the embodiment of the present invention.
FIG. 7 is an exemplary diagram for explaining a method for determining the degree of interest of a publicity target in an automatic learning system and method for deriving effective marketing according to an embodiment of the present invention; FIG.
8 is a flowchart showing a processing procedure of information stored and fed back to the learning storage unit 400 among the embodiments of the present invention.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

The automatic learning system and method for deriving effective marketing according to the embodiment of the present invention analyzes postings posted by users in SNS (Social Networking Service), grasps the propensity of the publisher and uses it for marketing, stores the analysis result It is about the system that includes the learning ability to derive optimal marketing because it is used as the comparative data in the next analysis.

The SNS is a service whose main purpose is to connect a plurality of users with each other and exchange information between users who have a relationship, and the specific service methods and terms are variously implemented according to the service provider. In the present description, the posting of the article, the image, the moving picture, and the link of the other web page is expressed as 'post' in the SNS, and the person who posted it is expressed as 'author'.

1 is a block diagram illustrating a configuration of an automatic learning system and method for deriving effective marketing according to an embodiment of the present invention.

As shown in the figure, an automatic learning system and method for deriving effective marketing for an embodiment of the present invention includes a collection unit 100 for searching a post including a search keyword from the SNS, and collecting information about the searched post information and the post author A reasoning unit 200 for selecting a publicity target among a plurality of creators collected by the collecting unit 100 and determining a publicity topic and a publicity method, a publicity topic determined by the publicity target selected by the reasoning unit 200, And a learning storage unit 400 for storing a response to the information to be promoted, a publicity topic, a publicity method, and a publicity .

It is preferable that each of the above-described configurations is not a physical entity but an application program that performs respective functions under an operating system of an electronic device such as a computer or a server. It is also to be understood that the respective configurations are not limited to being configured in a single electronic device but may be interconnected in an electronic manner such as a network after each configuration is applied to a plurality of electronic devices. However, in the case of the proof unit 300 that performs the actual promotion during the configuration, it may be an application program because the publicity should be performed in various ways to the selected publicity object corresponding to various publicity topics. However, It may be a kind of group consisting of a plurality of people in a company that conducts business on a basis and conducts public relations related tasks.

Here, the collecting unit 100 includes a keyword registration unit 110 for setting and registering a search keyword, an SNS collecting unit 120 for collecting post information by searching the SNS for a post including the search keyword, And an information extracting unit 130 for extracting necessary information and author information for deriving a publicity method from the information.

The reasoning unit 200 includes a target selecting unit 210 for grouping posters based on the post information collected by the collecting unit 100 and the information of the post creator, A channel selection unit 230 for comparing the individual information of the creator with the cumulative information stored in the learning storage unit 400 and deriving a publicity method based on the individual information of the user, .

The proof unit 300 includes an offer execution unit 310 for performing a publicity based on a publicity topic and a publicity method determined by the reasoning unit 200 to the publicity object selected by the reasoning unit 200, A result collecting unit 320 for collecting the information of a publicity object in response to the publicity information received from the public information collecting unit 320, ).

The overall system flow of the automatic learning system and the service of the method for deriving effective marketing according to the embodiment of the present invention will be described together with the constitution.

FIG. 2 is a message flow diagram illustrating a service process of an automatic learning system and method for deriving effective marketing according to an embodiment of the present invention. FIG. 3 is a flowchart illustrating an automatic learning system and method for deriving effective marketing according to an embodiment of the present invention Fig.

As shown, an automatic learning system and method for deriving effective marketing according to an embodiment of the present invention derives a publicity method in the following order.

The publicity method includes the steps of: (a) receiving a search keyword from the keyword registration unit 110 (S110); (b) sending a post including the search keyword received from the keyword registration unit 110 to the SNS collector unit 120 (C) a step (S130) of extracting necessary information and author information in order for the information extracting unit 130 to derive an optimal publicity method from the posts retrieved from the SNS collector 120; (d) a step S210 of grouping the post creators with the information extracted by the information extracting unit 130 and selecting a public information object among them, (e) the offer extracting unit 220 (F) selecting a publicity method suitable for the publicity target selected by the target selection unit 210, (f) selecting a publicity subject suitable for the publicity target selected by the target selection unit 210 (G) when the offer execution unit 310 receives the promotional note selected by the offer derivation unit 220 (H) a reaction collecting unit (320) for collecting information from the offer execution unit (310); a step (S310) of performing publicity to a publicity target selected by the target selection unit (210) according to a publicity method selected by the channel selection unit (I) the result processing unit 330 transmits the reaction information collected by the reaction collecting unit 320 to the learning storage unit 400, (S330).

When a user using the SNS registers a post on-line, the post is stored in the system of the SNS. When a company desiring to find an effective promotion (marketing) method among the environment where such a post is accumulated in the system of the SNS requests the person who implements the embodiment of the present invention to select a publicity method, If the keyword is provided together, it is executed from the above step (a).

Here, the search keyword or the keyword is preferably a word related to a subject to be promoted by the referral company and a sentence synthesized with the word, and an effective promotional method to be provided to the referral company is responsive to the subject to be promoted by the referral company This is because it is necessary to first select the publicity target that is expected to be performed, and then to conduct the publicity test on the object and derive the response from the publicity. In order to select a publicity target, a person who searches for a SNS including a keyword provided by a referral company, and selects an author of the post as a publicity target.

First, in step (a), a keyword received from the referral company is set and registered in the keyword registering unit 110 (S110). The keyword registering unit 110 is one of the configurations of the embodiment of the present invention. The setting and registration of the keyword registration can be easily performed on an input device such as a mouse or a keyboard to the contents displayed on the monitor in this embodiment, May be performed in such a manner that corresponding characters are input to the system according to an embodiment of the present invention. (b), the SNS collector 120 searches the posts stored in the SNS and collects a post including the keyword (S120). When the post is collected by the SNS collector 120, in step (c), the information extractor 130 extracts necessary information to derive an optimal PR method from the collected post, or extracts the author information (S130), the information of the individual of the author is obtained by a request to the provider operating the SNS or by collecting the information on the author's personal SNS page. Here, the necessary information may be information such as the date of creation of the post, the number of posts created during an arbitrary period, and the amount of the post. The 'arbitrary time period' may be set by a system administrator or a requesting company as in the past one month, for example, as a time reference for confirming whether a post author is interested in a field related to a search keyword. The author's personal information may also include the author's gender, age, residence, interests, and the number of other users involved.

4 is a flowchart for explaining driving of the target selection unit in the embodiment of the present invention.

Now, the personal information of the creator collected by the information extracting unit 130 may be stored in the target selecting unit 210 as similar or common characteristics (S210), and the group that is expected to be most responsive to the publicity among the groups is selected as the publicity target (S210).

If the flow of step (d) according to the target selection unit 210 is further specified. (D1) grouping the creators collected by the information extracting unit (210), (d2) having the same characteristics as the group created at the step (d1) of the past publicity targets stored in the learning storage unit (D3) if there is no group having the same characteristics as the group of the past publicity targets stored in the learning storage unit in the step (d2), it is checked whether there is a group set by the client in advance to be selected as a publicity target (D4) listing the group if there is a group identified in the step (d2) or (d3), and (d5) selecting the listed group as a final publicity target have.

In order to predict whether the group formed by the target selection unit 210 is good, it is necessary to determine whether or not the group having the same characteristics as the group created at the step d1 of the past PR objects stored in the learning storage unit 400 You can check whether you have a past public relations object. However, the presently created group may not be exactly the same as the features of the past public relations object stored in the learning storage unit 400 due to the difference in fashion and time. Therefore, 'the same feature' can not be regarded as completely matching all the features, but it can be seen that some features of the group are the same, so that the similarity is high or is included in the same range defined by the administrator.

If the learning storage unit 400 has a past public relations object having the same characteristics, the present group can be immediately added to the public relations target, or the public relations related information performed in the past public relations object, for example, It is possible to judge whether or not to select the current group as a publicity target considering the information such as the reaction rate.

If there is not one of the past promotion objects stored in the learning storage unit 400 that has the same characteristics as the current group, the information extraction unit 130 determines whether to select the current group as a publicity target Based on the received information.

However, it is possible that the client has already designated a publicity target to check the response rate. If there is an object already designated as described above, the characteristic of the publicity object already designated with the group created in the step (d1) You will be able to see. If the features coincide with each other, the designated promoting object includes the author of the post included in the current group.

5 is a flowchart for explaining the operation of the offer derivation unit in the embodiment of the present invention.

Next, in step (e), the offer derivation unit 220 selects a public relations topic suitable for the group selected by the target selection unit 210 (S220). (E1) the step (e1) of loading the target to be promoted in the target selection unit (S221), (e2) the step of (e2) (E3) whether a past public information object having the same characteristics as the public information object loaded by the target selection unit is stored in the learning storage unit, (E3) if the past public relations target corresponding to the public relations target is not stored in the learning storage unit, the learning storage unit may store the public relations subject in the environment corresponding to the current environment (E5) if the past public relations subject stored in the learning storage unit does not exist in the (e3) or (e4) step, the public relations subject matter is arbitrarily selected directly at step S227 It has a feature that proceeds.

It should be recalled that the embodiment of the present invention has the purpose of deriving an effective publicity method in the step of (e2) of the above-mentioned steps, In order to quickly identify effective public relations methods, the example finds public relations targets that can show a response that is greater than usual, and promotes them with public relations topics that the public relations target is interested in (S222). For this reason, it is possible to use the similarity of the search keyword provided by the referral company, but in order to elicit a greater reaction of the publicity target, the publicity subject corresponding to the common characteristic of the group is selected Or a publicity subject corresponding to the characteristics of the individual's personal information.

Next, in step (e3), it is checked whether the past PR object having the same characteristics as the PR object loaded by the target selection unit is stored in the learning storage unit 400. If it is already stored, the past PR object What is the subject of the publicity that was used to conduct the publicity, and what was the response rate of the publicity (S223).

If the corresponding publicity topic can not be found in step (e3), the past publicity topic performed in the environment corresponding to the current environment, i.e., weather, day, time, (S225). ≪ / RTI >

If the publicity topic is not found in step (e4), the client or the system administrator directly selects the publicity topic as step (e5) (S227).

If the past public relations topic corresponding to the learning storage unit 400 is stored in the step (e3) or (e4), a step of selecting any one of the public relations topics may be performed (S229).

6 is a flowchart for explaining the operation of the channel selection unit in the embodiment of the present invention.

The step (f), which is the driving order of the channel selection unit 230, is performed as follows. The channel selection unit 230 selects (f1) the target to be promoted in the target selection unit and the offer derivation unit A step S232 of loading a publicity subject, (f2) a past publicity method performed on the past publicity object if the past publicity object having the same characteristics as the publicity object is stored in the learning storage, (f3) a step (S23) of loading a past PR method performed on the past PR subject if a past PR subject having the same characteristics as the PR subject is stored in the learning storage, (f4) If there is a publicity method loaded in the step (S235), (f5) the method having the best response rate among the publicity methods sorted in the step (f4) is selected, Combined PR It has a feature that goes to step 236 to generate the law.

Here, the public relations method refers to the kind of publicity to be performed for the public relations target. The types of public relations include a comment inputting an article added to the SNS post, a note supported by the SNS system, E-mail, SMS, TM (telemarketing) Direct marketing).

If there is no past publicity method to be loaded in the learning storage unit 400 in step (f2) or (f3) (S234), the step of the channel selection unit 230 ends, When the offer execution unit 310 recognizes that there is no method, the offer execution unit 310 selects an arbitrary publicity method itself or causes the requesting company to select a publicity method and executes the publicity. However, the publicity subject currently created in step (f3) may not be exactly the same as the characteristics of the past public relations topic stored in the learning storage unit 400 due to the difference in the fashion and the time as in the channel selection unit 230 . Therefore, 'same' can not be regarded as completely matching all features, but it can be seen that some characteristics of a public relations topic are the same, so they are highly similar, are related to each other, or are included in the same range defined by the manager.

Combining the different types of public relations methods in the step (f5) is to combine two or more public relations methods that show a better response rate than those of the past public relations methods that are stored in the learning storage unit 400 If the past multiple PR method is stored when the PR is generated by combining the different types of PR methods with the learning storage unit 400 and the response rate is higher than when PR is performed alone, It may be used again as a combined PR (S236). In order to combine the above-mentioned multiple PR or different types of PR methods, in addition to the publicity method using the SNS, a publicity method such as a TM for directly dialing, a DM for sending a mail, and an SMS for transmitting text via a mobile communication network A form that combines at least two public relations methods.

However, among these public relations methods, TM that sends a direct call to a public relations person or DM that sends a mail may have a higher execution cost than other public relations methods, so that the execution cost of the derived public relations method is higher than the cost paid by the client A large case may occur (S237). When such a situation occurs, the channel selection unit 230 may change the publicity method that can be executed more inexpensively or may reduce the number of publicity targets (S239).

When the derivation of the publicity method of the channel selection unit 230 is completed in step (f), the offer execution unit 310 performs publicity based on the publicity target and the publicity topic selected as the step (g) and the derived publicity method (S310).

FIG. 7 is an exemplary diagram for explaining a method of determining the degree of interest of a publicity target in an automatic learning system and method for deriving effective marketing according to an embodiment of the present invention. FIG. Which is stored in the storage unit 400 and fed back.

The information extracting unit 130 can count the number of times the author has created a post including the keyword when searching the post containing the search keyword in the SNS and collecting the information of the author who created the post. The authors who have created a large number of postings containing the keyword can judge that they are interested in the field related to the keyword. Therefore, collecting the number of related posts of each author together with other information, Will be selected.

However, the author may have a negative interest in the field related to the keyword, but in such a case, it would be preferable to exclude it from the publicity target. Therefore, when searching for a post including a keyword, in order to acquire as many post creators as possible, a plurality of similar keywords having a meaning similar to a search keyword and a keyword are searched for, and if an author having a post with the keyword is searched, , The search is further performed to find out whether expressions such as "no", "bad", "return", "useless" are included together. Posts that do not use negative keywords, including search keywords, will be rated as high interest by the author, but if the search keywords and negative expressions are included in the post, the author is considered to be less interested and may be excluded There will be.

In addition, when collecting the creator information, the information extraction unit 130 may collect the number of users related to the creator together in the SNS, and evaluate the influence of each creator. Users who are related to the author within the SNS can be other SNS users who are registered as authors, friends or neighbors and mutually share the information of each other. Even if they are not registered as friends or neighbors, Other SNS users who are using the SNS can also be seen as users related to the author.

Since authors with a large number of related users are more likely to spread the information contained in the posts than other SNS users even if they post a single post, authors having a certain number or more of related users may be selected for publicity .

If there is a group having the same characteristics as the past public relations target stored in the learning storage unit 400, the target selection unit 210 may select the group immediately as a publicity target or determine whether to select the group as a publicity target And if there is nothing that has the same characteristics as those of the group of the past publicity targets of the learning storage unit 400, the publicity information is added to the publicity information that has already been held, or only the information received from the information extraction unit 130 It can be judged whether or not to select it as a target.

The offer derivation unit 220 reflects the response rate at the time of performing the publicity by using the publicity subject if the publicity subject performed in the current environment is already stored in the learning storage unit 400 when the publicity subject is selected Promotion subject can be selected. If the response rate of the PR performed on the past PR topic stored in the learning storage unit 400 is lower than the average response rate of the other PR topics, the past PR topic will be excluded from the PR topic, If it is high, it can be selected as a publicity topic again.

In addition, the offer execution unit 310 can perform publicity by placing a difference in environment such as day of the week, time of day, weather, and the like when executing the publicity. If the publicity target is divided into a certain number of people and the publicity is divided by a certain number of people, the response rate will be different even if the target group has similar characteristics. If so, the same publicity target, , It can be used as a reference material for the next publicity.

In addition, the offer execution unit 310 must perform publicity by various publicity methods. However, publicity methods such as TM or DM may be limited to fully automation by electronic devices. Therefore, all or a part of the offer executor 310 may receive a publicity object and a publicity method derived from the speculation unit 200, and then receive publicity information, which is composed of a plurality of persons, Group.

However, in the case of a DM, the actual address of the author's residence or workplace is required. Since such personal information may be very difficult to collect, ask the SNS administrator who can inquire the personal information to request DM to execute the DM It may be possible to proceed.

The response collecting unit 320 collects the response time information, the time it takes for the response to come and the response after the publicity run, the current weather, the day of the week, etc. Environmental information of the environment. In response to the publicity, when a publicity target is clicked on a link or a URL included in the publicity content such as a comment, a note, or an e-mail, the response is collected by the response collecting unit 320 In the case of promoting by TM, it can be treated as responding when the publicity object responds favorably. In the case of promoting by DM, the publicity object is tried to be accessed by the URL included in the letter , The QR code enclosed in the letter can be scanned by the corresponding scanner, and the corresponding information can be handled as a response.

In addition, the offer execution unit 310 can simultaneously perform the publicity by using two or more publicity methods to one publicity target when the publicity is executed. For example, the same promotional topic is to be promoted as a text message (SMS) with a comment. When the compound advertisement method is executed, the reaction collecting unit 320 records the time and the response rate until a response is made in response to the public information, It is possible to obtain the degree of synergy that occurs when a public relations method is combined.

Since the reaction collecting unit 320 can not wait indefinitely for the response sent by the publicity targets, it is possible to limit the time for waiting for the reaction. When the reaction collecting unit 320 reaches the limit time of the reaction collecting, The processing unit 330 arranges it and transmits it to the learning storage unit 400. In order to organize information, the result processing unit 330 groups the information of the authors constituting the publicity target, the publicity topic, the publicity method, the publicity execution environment, and the environmental information when the reaction is collected into one result information, (S330). ≪ / RTI >

At the same time, data on the public relations result sorted by the result processing unit 330 is also provided to the requesting company.

The learning storage unit 400 stores and stores the information transmitted by the result processing unit 330 and is used as a comparison data in the process of deriving a publicity method to be performed thereafter. Further, based on the data, You can also analyze trends.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is to be understood that the invention is not limited to the disclosed exemplary embodiments. It is clear that the present invention can be suitably modified and applied in the same manner. Therefore, the above description does not limit the scope of the present invention, which is defined by the limitations of the following claims.

100: collecting unit 110:
120: SNS collector 130: Information extractor
200: Reasoning unit 210:
220: an offer derivation unit 230: a channel selection unit
300: Proof unit 310: Offer execution unit
320: reaction collecting unit 330: result processing unit
400: learning storage unit

Claims (17)

A collecting unit 100 for searching a post including a search keyword in the SNS, and collecting information about the searched post information and the post author,
A reasoning unit 200 for selecting a publicity target based on the information collected by the collecting unit 100 and determining a publicity topic and a publicity method,
A proof unit (300) for performing a publicity by a publicity subject and a publicity method determined by the publicity object selected by the reasoning unit (200) and collecting the publicity target,
And a learning storage unit (400) for storing the information to be publicized, the publicity topic, the publicity method, and the publicity response. The apparatus according to claim 1, wherein the collecting unit (100)
A keyword registration unit 110 for setting and registering a search keyword, an SNS collection unit 120 for collecting post information by searching a post including the search keyword in the SNS, And an information extraction unit (130) for extracting information and author information. The apparatus according to claim 1, wherein the reasoning unit (200)
A target selecting unit 210 for grouping posters based on the post information collected by the collecting unit 100 and the information of the post creator and selecting an object to be publicized among them, And a channel selection unit 230 for comparing the individual information of the creator with the cumulative information stored in the learning storage unit 400 and deriving a publicity method. An automatic learning system that leads to marketing. The apparatus according to claim 1, wherein the proof unit (300)
An offer execution unit 310 that executes a publicity based on a publicity topic and a publicity method determined by the reasoning unit 200 to the publicity target selected by the reasoning unit 200, A response collecting unit 320 for collecting information of a publicity target and a result processing unit 330 for communicating the information of the publicity information collected by the reaction collecting unit 320 to the learning storage unit 400 An automatic learning system that leads to effective marketing. A keyword registration unit 110 for setting and registering a search keyword,
An SNS collector 120 for collecting post information by searching the SNS for a post including the search keyword,
An information extraction unit 130 for extracting information and author information necessary for deriving a publicity method from the retrieved post information,
A target selecting unit 210 for grouping posters based on the post information extracted by the information extracting unit 130 and the information of the post creator,
An offer derivation unit 220 for selecting a publicity subject based on the post information extracted by the information extraction unit 130 and the information of the post creator,
A channel selecting unit 230 for comparing the post information extracted by the information extracting unit 130 and the information of the post creator with the accumulated information stored in the learning storage unit 400 to derive a publicity method,
An offer execution unit 310 for performing promotion by a publicity method in which the channel selection unit 230 derives a publicity topic selected by the offer derivation unit 220 to a publicity target selected by the target selection unit 210,
A reaction collecting unit 320 for collecting publicity information in response to the publicity performed by the offer executing unit 310,
A result processing unit 330 for transmitting the information to be promoted collected by the reaction collecting unit 320 to the learning storage unit 400,
And a learning storage unit (400) for storing the information to be promoted, a publicity subject, a publicity method, and a publicity response. 6. The method according to claim 2 or 5,
The information extracting unit 130 evaluates the interest on the basis of at least one of the creation date of the post, the number of posts created during an arbitrary period, and the creation amount when collecting the post information including the search keyword and the information of the post creator An automatic learning system that derives effective marketing features. The method according to claim 6,
The information extracting unit 130 calculates an influence degree based on the number of other users related to the creator in the SNS when collecting the post information including the search keyword and the information of the post creator, An automatic learning system. The method according to claim 3 or 5,
When there is a group having the same characteristics as the past public relations target stored in the learning storage unit 400, the target selection unit 210 determines whether the group is to be immediately added to the publicity target or the group is to be selected as the publicity target A point of addition is given,
And determines whether to select a publicity target based on only the information received from the information extracting unit 130 when there is no one having the same characteristics as the past publicity targets of the learning storage unit 400 An automatic learning system. The method according to claim 3 or 5,
If the public relations subject performed in the current environment has already been stored in the learning storage unit 400, the offer derivation unit 220 reflects the reaction rate at the time of performing the public relations promotion An automatic learning system that derives an effective marketing feature that selects a public relations topic. The method according to claim 4 or 5,
Wherein the offer execution unit (310) executes a publicity by placing a difference in environment or time at the time of executing the publicity. 11. The method of claim 10,
The response collecting unit 320 collects the response time, the time until the response after the publicity run, and the environment information when the publicity target is responded to the publicity performed by the offer execution unit 310 An automatic learning system that leads to effective marketing. The method according to claim 4 or 5,
Wherein the offer execution unit (310) promotes one public relations topic by combining two or more public relations methods in the same environment or time when executing public relations. 13. The method of claim 12,
The response collecting unit 320 calculates the degree of increase of the reaction rate more than the single publicity method when the publicity target is responded to the combined publicity information executed by the offer execution unit 310 to combine the different publicity methods The degree of synergy that occurs in the case where the marketing is performed. (a) setting and registering a search keyword by the keyword registering unit 110;
(b) retrieving, from the SNS, a post including the search keyword received by the SNS collector 120 from the keyword registering unit 110;
(c) extracting necessary information and author information in order for the information extracting unit 130 to derive a publicity method from the posts retrieved from the SNS collector 120;
(d) grouping the post creators with the information extracted by the information extracting unit 130 by the target selecting unit 210, and selecting a public information target;
(e) selecting a publicity subject based on individual information of the post creator extracted by the offer derivation unit 220 from the information extraction unit 130;
(f) deriving a publicity method by comparing the individual information of the creator with the cumulative information stored in the learning storage unit (400) by the channel selection unit (230);
(g) The offer execution unit 310 executes a publicity target selected by the target selection unit 210 using the publicity method selected by the channel selection unit 230, based on the publicity topic selected by the offer derivation unit 220 step;
(h) collecting the reaction information of the publicity object in response to the publicity performed by the reaction executing section (310) by the reaction collecting section (320);
(i) transferring the reaction information collected by the reaction collecting unit 320 to the learning storage unit 400 by the result processing unit 330;
(j) storing the information transmitted from the result storage unit (330) by the learning storage unit (400). 15. The method of claim 14, wherein step (d) comprises:
(d1) grouping the creators collected by the information extracting unit 130;
(d2) confirming whether the group has the same characteristics as the group created at the step (d1) of the past publicity targets stored in the learning storage unit (400);
(d3) checking whether there is a group set in advance by the client to be selected as a publicity target in the case where there is no characteristic that has the same characteristics as the past publicity target group stored in the learning storage unit 400 in the step (d2) ;
(d4) if the group identified in the step (d2) or (d3) exists, cataloging the group;
(d5) selecting the listed group as a final publicity target. 15. The method of claim 14, wherein step (e) comprises:
(e1) loading the publicity target in the target selection unit 210;
(e2) selecting a feature to be a reference for selecting a publicity topic among the publicity target information;
(e3) confirming whether or not a past PR object having the same characteristics as the PR object loaded by the target selection unit 210 is stored in the learning storage unit 400, a used PR topic, and a reaction rate;
(e4) If the past PR object corresponding to the PR object is not stored in the learning storage unit 400 in step (e3), the PR topic performed in the environment corresponding to the current environment is stored in the learning storage unit 400 ;
(e5) a step of arbitrarily directly selecting a publicity topic if there is no past public relations topic stored in the learning storage unit (400) in the step (e3) or (e4) . 15. The method of claim 14, wherein step (f) comprises:
(f1) loading the publicity subject and the publicity subject in the target selection unit 210 and the offer derivation unit 220;
(f2) loading a past public relations method performed to the past public relations object if a past public relations object having the same characteristics as the public relations object is stored in the learning storage unit (400);
(f3) loading a past publicity method performed on the past publicity topic if the past publicity topic having the same characteristics as the publicity topic is stored in the learning storage unit 400;
(f4) if there is a publicity method loaded in the step (f2) or (f3), arranging in the order of good response rate;
(f5) a step of selecting a method having the best response rate among the advertisement methods sorted in the step (f4), and generating a compound advertisement method by combining different kinds of advertisement methods Learning method.

Images